One of the indices of medical treatment for diabetes mellitus that are known and currently being used is the amounts of glycosylated hemoglobins (i.e., hemoglobin A.sub.1), and the amount of hemoglobin A.sub.1c in particular. As is shown below in equation 1, hemoglobin A.sub.1c is a complex in which one molecule of glucose is bound non-enzymatically to the N-terminal amino acid, valine, of the .beta.-chain of hemoglobin (below, abbreviated Hb). When the glucose has been bound to the amino acid mentioned above, first, as shown below in the equation, labile HbA.sub.1c I, which is a Schiff base, is formed: ##STR1##
In the above equation 1, .beta.--A--NH.sub.2 represents Hb, and the NH.sub.2 here shows the amino group in the N-terminal amino acid, valine, of the .beta.-chain of the said Hb.
The reaction that produces this labile HbA.sub.1c I, is a reversible reaction, and depending on the glucose concentration, the equilibrium will tend to go either in the direction of the production of the labile HbA.sub.1c I, or else it will tend to go in the direction of its elimination. Compound I is converted irreversibly into the stable HbA.sub.1c II by the Amadori rearrangement.
HbA.sub.1c can be separated from other hemoglobins by means of high-performance liquid chromatography, and measured in terms of its optical density (OD), but it is not possible to separate the stable HbA.sub.1c (S--HbA.sub.1c) and the labile HbA.sub.1c (L--HbA.sub.1c) and to measure them separately. Therefore, it is not possible to obtain a reliable value of HbA.sub.1c. The reason is that the amount of L--HbA.sub.1c varies depending on the concentration of glucose (that is, the blood glucose) that is present, and the concentration of glucose in the blood changes rapidly and greatly, depending on meals and physical activity.
To solve the problems described above, attempts to eliminate L--HbA.sub.1c from blood samples and to measure the amount of S--HbA.sub.1c have been made. For example, in a paper by David M. Nathan et al. (Clinical Chemistry, 28, 512-515, 1982 ), semicarbazide and aniline are used as agents to eliminate the LH--HbA.sub.1c, and it is disclosed that the blood sample is treated with these agents at 38.degree. C. for 30 minutes. The semicarbazide captures the glucose, and acts as a nucleophilic reagent as well, competing with the amino group of the Hb. The aniline acts as a catalyst. The result is that virtually all of the L--HbA.sub.1c is eliminated. However, because the reaction for the elimination of L--HbA.sub.1c is at an acidic pH (pH 5.0) and at a relatively high temperature (38.degree. C.) for a long period of time (30 minutes), some denaturation of the Hb (for example, elimination of the heme) may occur. For example, when the elution pattern for ion-exchange chromatography is examined, the height of the peak decreases owing to fading of color caused by the elimination of heme, and the peaks that correspond to HbA.sub.1a and HbA.sub.1b respectively are seen as being larger.
In Japanese Laid-Open Publication 58-210024, a dihydroxyboryl compound (i.e., a derivative of boric acid) that is as an agent for the elimination of LH--HbA.sub.1c is disclosed. This dihydroxyboryl compound reacts with glucose to form a complex, the result of which is to cause the elimination of the L--HbA.sub.1c, However, to eliminate the L--HbA.sub.1c, high concentrations of the dihydroxyboryl compound are needed. For example, about 0.1-1.0M of the said compound is needed for a sample that contains blood that has been hemolyzed. When the hemolyzed blood is put on the column with the use of an eluent that contains the said dihydroxyboryl compound, it is necessary to use the compound at the concentrations of 0.01-0.15M in the eluent. When this kind of high concentration of dihydroxyboryl compound is used for the eluent for ion-exchange chromatography, an ionic strength of the eluent is different from that of ordinary eluents. The result is that because the separation conditions are changed, measurement can become difficult, or it is necessary to make changes in the measurement conditions. Furthermore, because the optimum pH for the formation of the complex mentioned above is about 4.5-6.5, and preferably 5.0-6.0, there is a danger that the Hb will be denaturated.
As another method for the elimination of L-HbA.sub.1c, there is a method in which the blood sample is diluted, thereby lowering the glucose concentration, resulting in an acceleration of elimination of L--HbA.sub.1c. When this method is employed practically, for example, erythrocytes are incubated in a large excess of physiological saline, or the hemolysate is dialyzed. However, all of these processes require a long period of time, and thus they are not appropriate methods for use in clinical testing.